The nonsmooth multiobjective fractional robust optimization problem is investigated. By employing the Gordan alternative theorem and minimax theorem, necessary optimality conditions are established under nonsmooth Extended Mangasarian-Fromovitz constraint qualifications. Subsequently, generalized convex-concave functions are introduced. A Mond-Weir type dual model is then proposed. Under generalized convexity assumptions, strong duality, weak duality, and converse duality theorems between the dual problem and the original problem are derived. These results enrich the theoretical framework of optimization theory and provide new algorithmic foundations for solving multiobjective optimization problems.